aitune.torch.tune_strategy.mixin.performance_validation_mixin

Performance validation mixin for tune strategy.

Module Contents

Classes

Name	Description
`PerformanceValidationMixin`	TuneStrategy mixin that validates each backend against a TorchEager throughput baseline.
`PerformanceValidationMixinResult`	Throughput and speedup result for a single backend.

Functions

Name	Description
`fmt_speedup_msg`	Returns the full speedup summary line with module/backend fields.
`fmt_speedup_msg_short`	Returns a compact speedup line without module/backend fields.

API

class aitune.torch.tune_strategy.mixin.performance_validation_mixin.PerformanceValidationMixin(
    args = (),
    min_speedup_threshold_percent: float = 1.0,
    kwargs = {}
)

Bases: TuneStrategy

TuneStrategy mixin that validates each backend against a TorchEager throughput baseline.

When performance validation is enabled, profiles TorchEager during _pre_tune with the strategy profiling config narrowed to the resolved batch size (from graph_spec.get_max_batch_size()) to establish a baseline. For every candidate backend, profiles with the same task-local profiling config and appends a PerformanceValidationMixinResult. Backends with speedup below the configured threshold are rejected.

_baseline_backend

Backend | None = None

_baseline_throughput

float | None = None

_performance_validation_enabled

bool = True

_resolved_batch_size

int | None = None

perf_validation_results

list[PerformanceValidationMixinResult] = []

aitune.torch.tune_strategy.mixin.performance_validation_mixin.PerformanceValidationMixin._build_validate_and_check_perf(
    backend: aitune.torch.backend.backend.Backend,
    module: torch.nn.Module,
    name: str,
    graph_spec: aitune.torch.module.graph_spec.GraphSpec,
    data: list[aitune.torch.module.recording_module.Sample],
    device: torch.device,
    cache_dir: pathlib.Path,
    raise_on_failure: bool = False
) -> aitune.torch.backend.backend.Backend | None

Build, correctness-validate, then profile and check throughput against baseline.

Returns the built backend, or None when correctness fails (raise_on_failure=False) or performance check rejects it (speedup below threshold).

aitune.torch.tune_strategy.mixin.performance_validation_mixin.PerformanceValidationMixin._log_baseline_selected(
    backend: aitune.torch.backend.backend.Backend
) -> None

Emit an explicit message when the TorchEager baseline is the selected backend.

aitune.torch.tune_strategy.mixin.performance_validation_mixin.PerformanceValidationMixin._post_tune(
    backend: aitune.torch.backend.backend.Backend | None,
    name: str,
    graph_spec: aitune.torch.module.graph_spec.GraphSpec,
    data: list[aitune.torch.module.recording_module.Sample]
)

Emits a speedup line after tuning completes.

aitune.torch.tune_strategy.mixin.performance_validation_mixin.PerformanceValidationMixin._pre_tune(
    module: torch.nn.Module,
    name: str,
    graph_spec: aitune.torch.module.graph_spec.GraphSpec,
    data: list[aitune.torch.module.recording_module.Sample],
    device: torch.device,
    cache_dir: pathlib.Path
)

Runs pre-tune setup and profiles TorchEager when performance validation is enabled.

aitune.torch.tune_strategy.mixin.performance_validation_mixin.PerformanceValidationMixin._profiling_config_for_batch_size(
    batch_size: int,
    batching: bool | None = None
) -> aitune.torch.task.profiling.ProfilingConfig

Returns the strategy profiling config narrowed to one validation batch size.

aitune.torch.tune_strategy.mixin.performance_validation_mixin.PerformanceValidationMixin.enable_performance_validation(
    enable: bool = True
) -> aitune.torch.tune_strategy.mixin.performance_validation_mixin.PerformanceValidationMixin

Enables or disables TorchEager baseline profiling and candidate performance checks.

class aitune.torch.tune_strategy.mixin.performance_validation_mixin.PerformanceValidationMixinResult(
    backend_description: str,
    throughput: float,
    baseline_throughput: float,
    speedup: float,
    passed: bool
)

Dataclass

Throughput and speedup result for a single backend.

backend_description

str

baseline_throughput

float

passed

bool

speedup

float

throughput

float

aitune.torch.tune_strategy.mixin.performance_validation_mixin.fmt_speedup_msg(
    speedup: float,
    detail: str,
    name: str,
    backend_desc: str
) -> str

Returns the full speedup summary line with module/backend fields.

aitune.torch.tune_strategy.mixin.performance_validation_mixin.fmt_speedup_msg_short(
    speedup: float,
    detail: str
) -> str

Returns a compact speedup line without module/backend fields.